우리나라 해운기업의 부실예측에 관한 연구

A Study on the Failing Prediction of Korean Shipping Companies

The purpose of this paper is to make the prediction models of business failure using multivariate discriminant analysis and logit analysis for Korean shipping compames. Many shipping compames have gone out of business and we find a list of companies closed a business between 2000 and 2014 in a report of cessation of business from the National Tax Service. Only 32 firms of all failure shipping companies were available to collect the financial statement in DART(Data Analysis, Retrieval and Transfer System) of the Financial Supervisory Service. Finally the 32 firms became a sample group for this study. And another normal 32 firms were chosen as a control group for comparison with failure firms. We considered 22 financial ratios of 5 categories for independent variables. And we selected independent variables by the result of t-test between two groups. Also the 32 paired firms were divided into two groups. One is to fit the prediction model and the other group is to test reliability of the fitting model. We assessed the accuracy in classification by hit ratio and reliability by predictive power. The results of this study are as follows. First, when using a year ago data from closing business, the hit ratio of two models is the same with 79.5%. However the predictive power of logit model with 75% is higher than discriminant model with 70%. Second, when using data of two-years ago, the hit ratio of discriminant model is 65 .0% and the predictive power is 66.7%. However the significance of discriminant model is not valid in the 95 % confidence level. In case of logit model, the hit ratio is 55.5% and the predictive power is 50%. The significance of logit model is valid in the 95 % confidence level however t-values of all independent variables are not significant.